CARBON DIOXIDE CAPTURE BY TERTIARY AMIDINE FUNCTIONAL ADSORBENTS. Stephen A. Gattuso. B.S. Chemistry, Duquesne University, 2003

Transcription

1 CARBON DIOXIDE CAPTURE BY TERTIARY AMIDINE FUNCTIONAL ADSORBENTS by Stephen A. Gattuso B.S. Chemistry, Duquesne University, 2003 Submitted to the Graduate Faculty of School of Engineering in partial fulfillment of the requirements for the degree of Master of Science University of Pittsburgh 2007

2 UNIVERSITY OF PITTSBURGH SCHOOL OF ENGINEERING This thesis was presented by Stephen A. Gattuso It was defended on April 06, 2007 and approved by Dr. Irving Wender, Distinguished University Research Professor, Department of Chemical and Petroleum Engineering Dr. Götz Veser, Assistant Professor, Department of Chemical and Petroleum Engineering Dr. Toby Chapman, Associate Professor, Department of Chemistry Thesis Advisor: Dr. Robert M. Enick, Professor and Chairman, Department of Chemical and Petroleum Engineering ii

3 Copyright by Stephen A. Gattuso 2007 iii

4 CARBON DIOXIDE CAPTURE BY TERTIARY AMIDINE FUNCTIONAL ADSORBENTS Stephen A. Gattuso, M.S. University of Pittsburgh, 2007 In the recent past, carbon dioxide capture from gas streams has been done using multiple methods. These methods include cryogenic, membrane, O 2 /CO 2 recycle combustion systems, physical absorption, and chemical absorption. The most common of these methods is chemical absorption, which typically uses primary and secondary amines. MEA and DEA are the most common amines used for carbon dioxide capture. These amines bind carbon dioxide at a 2:1 molar ratio. These systems are liquid systems that require large amounts of energy for regeneration and recirculation. Liquid amine systems typically bind 2.5 to 4 mol CO 2 /kg absorbent at the 2:1 molar ratio. Liquid tertiary amines and amidines have been shown to bind carbon dioxide at a 1:1 molar ratio, thereby reducing the volume of amine, but the binding rate is much slower. Amine blends have been used to compensate for this difference. In order to reduce regeneration and recirculation costs and increase CO 2 binding capacity, solid adsorbents are being considered. Solid adsorbents using amidines allow for low regeneration costs and twice the carbon dioxide capacity per mol of adsorbent. This research focuses on the creation of amidine functionalized solid adsorbents to bind carbon dioxide in the presence of water. Amidine and guanidine functional groups increase capacity for carbon dioxide binding. Several amidine compounds and a guanidine containing compound have been synthesized and show increased binding capacity of carbon dioxide compared to conventional liquid systems. By reducing molecular weight of the nonbinding portion of the polymer, an amidine polymer can iv

5 bind almost two times that of the other created adsorbents. The polyamidine bound 9.30 mol CO 2 /kg polymer while the polyguanidine bound over 6 mol CO 2 /kg polymer. These experiments were done in a 10 ml batch reactor at 45 psi at room temperature. Water (liquid) was added prior to the experiment at a 1:1 molar ratio with the binding sites in the polymer. There is still much work to be done to understand amidine polymer binding completely, including kinetic tests, but there is much promise in amidine polymer adsorbent technology. v

6 TABLE OF CONTENTS PREFACE...XI 1.0 INTRODUCTION CARBON DIOXIDE STORAGE CARBON DIOXIDE CAPTURE BACKGROUND CRYOGENIC CAPTURE MEMBRANE CAPTURE O 2 /CO 2 RECYCLE COMBUSTION SYSTEM CAPTURE PHYSICAL ADSORPTION CHEMICAL ABSORPTION HYPOTHESIS BINDING EFFICIENCY STABILITY SPECIFIC PROJECT AIMS EXPERIMENTAL PREPARATION OF 1,8-DIAZABICYCLO[5.4.0]UNDEC-7-ENE (DBU) DBU SILICA p-chloromethylphenyl silica vi

7 4.2.2 DBU-methylphenyl silica DBUDODECANE ,4-DIDBUBUTANE POLYETHER DBU POLYGUANIDINE POLYAMIDINE CO 2 ADSORPTION EXPERIMENTAL RESULTS DBU-METHYLPHENYL SILICA DBUDODECANE ,4-DIDBUBUTANE POLYETHER DBU POLYGUANIDINE POLYAMIDINE CONCLUSIONS FUTURE WORK STRUCTURE CONSIDERATIONS BINDING AND KINETIC CONSIDERATIONS APPENDIX A BIBLIOGRAPHY vii

8 LIST OF TABLES Table 1: CO 2 purity and recovery for membrane separation systems at 16 bar Table 2: Structures and pka values of common amines Table 3: pka Values of Amidine and Guanidine Functional Compounds. *pka of guanidine Table 4: Polyamidine CO 2 Capture in Batch Reactor Table 5: CO 2 Capture by Product viii

9 LIST OF FIGURES Figure 1: Primary and secondary amine systems bind CO 2 at a 2:1 ratio... 9 Figure 2: Hindered amines, such as 2-amino-2-methyl-1-propanol, bind at a 1:1 ratio Figure 3: DBU binds CO 2 at a 1:1 ratio. No carbamate is formed by this mechanism Figure 4: Imines are unstable while amidine and guanidine groups are stable Figure 5: Proton removal on DBU by BuLi Figure 6: Schematic of the DBU-methylphenyl silica synthesis Figure 7: 1-DBUdodecane reaction scheme Figure 8: Polyether DBU reaction scheme. DBU forms a bulky side chain in the polyether Figure 9: Polyguanidine reaction scheme Figure 10: Polyamidine reaction scheme Figure 11: TGA of DBU-methylphenyl silica. CO2 weight loss 8.3% Figure 12: DBU-methylphenyl silica (three cycles). Additional weight loss from degradation. 33 Figure 13: NMR of 1-DBUdodecane in CdCl 3. Peak at 1.23 indicates chain CH 2 protons Figure 14: 1 H-NMR of 1,4-diDBUbutane in CdCl Figure 15: NMR of the polyether DBU. CdCl3 peak occurs at 7.24 ppm Figure 16: 1 H-NMR of polyguanidine Figure 17: NMR of the solid amidine polymer ix

10 Figure 18: Liquid polyamidine NMR due to shorter chain lengths x

11 PREFACE I would like to thank Dr. Bob Enick for his assistance in completing my thesis and for stepping in as my thesis advisor. I really appreciate everything you have done for me. I would like to thank my research advisor Dr. Beckman for his guidance and direction as well as NETL and the DOE for funding this research. I would like to dedicate this to my mother and father, for shaping me into the person who I am today. A special thank you goes to my grandparents for their support. Thank you to my brothers and the rest of my family for all their support and guidance through these years. You have all made valuable additions to my education over the years. Thank you all. xi

12 1.0 INTRODUCTION Environmental issues are a major cause for concern in the world of today. Greenhouse gases are an environmental concern that is continuously increasing in importance, especially the emissions of carbon dioxide. Fossil fuel burning results in the emission of carbon dioxide. This additional carbon dioxide is upsetting the delicate balance that the earth manages naturally. There are too many sources of anthropogenic carbon dioxide adding to the concentration of CO 2 in the atmosphere every year, and these sources are only increasing as the world becomes more industrialized. Recently industries have been taking more proactive steps to remove carbon dioxide from their emission gas streams. Global carbon dioxide emissions exceeded 25 billion metric tons in The United States alone accounted for 23.1% of the total carbon dioxide emissions. The next closest carbon dioxide emitter was China at only 14.1% of the total carbon dioxide emissions. The United States emissions were 5.8 billion metric tons of carbon dioxide which is roughly equal to the emissions of the United Kingdom, Canada, Russia, Germany, and Japan combined. 1 The most abundant emission source from power plants in the United States is from coal-based power plants. Coal plants account for 80% of the total carbon dioxide emissions from fossil fuels. 2 1

13 1.1 CARBON DIOXIDE STORAGE In order to reduce the amount of carbon dioxide released into the atmosphere, the carbon dioxide must be captured and stored, or sequestered. With the vast amounts of carbon dioxide being generated there must also be very large amounts of storage areas to hold the gas. Some options which seem potentially viable include ocean storage, geological storage, and conversion into inorganic carbonates. 3 Carbon dioxide is absorbed by the ocean allowing for the possibility to pump excess carbon dioxide into the deep ocean. The ocean has a large storage capacity for the carbon dioxide, but there are possible concerns with this technique. The addition of large amounts of carbon dioxide may change the ph of the ocean where it is being absorbed, which could alter the ecology of that area of ocean. Due to this potential hazard and political considerations, this does not lend itself as the best choice for CO 2 storage. Carbon dioxide can also be stored in porous geological formations such as deep saline aquifiers, depleted oil and gas beds, and unmineable coal beds. Storage of carbon dioxide in coal beds that are too deep or too thin to mine is based on the ability of the coal to adsorb the CO 2 that is pumped into it. The injection of CO 2 may swell the coal, however, resulting in reduced CO 2 integrity. Saline aquifiers consist of areas deep in the earth that are filled with water that has high concentrations of salts. These are not useable for human use and would hold a large amount of carbon dioxide. Another intriguing storage option is oil and gas beds whose natural resources are spent. Adding CO 2 provides a mechanism to store the CO 2 and also drive out more oil or natural gas that could not have been mined before. This type of storage mechanism reduces the costs to store as it creates a new profit from a depleted source. 2

14 1.2 CARBON DIOXIDE CAPTURE In order to store carbon dioxide, the gas must be captured prior to release into the atmosphere. Currently CO 2 capture is very energy expensive. A coal-oil mixture plant suffered a penalty of 27% in efficiency when an amine-based carbon dioxide capture system was implemented. 4,5 This power reduction was 0.4 kwh/kg CO 2 captured. 4 The capture requires additional power usage by the specific plant, which reduces the plant s efficiency. Therefore in order to make carbon dioxide capture economically feasible, the system must be as efficient in capturing carbon dioxide as possible for a minimum installation cost. Another economic impact of CO 2 capture is the cost of regeneration of the solvent. Capture using amine systems can result in additional costs up to 77 U.S. dollars/ton CO 2 captured. 6 The majority of the additional cost occurs from heating the CO 2 rich solvent (absorbing liquid) for regeneration at high temperatures and low pressures. These liquid systems, which are mostly aqueous-mea based, have relatively large heat capacities which require a large amount of heat to release the CO 2. 7 This is due to the high heat capacity of water as well as MEA. A typical MEA process can use up to 37% of the total energy output of a power plant. 8 Up to 80% of the energy costs are due to the steam required to regenerate the absorbent. 6 Energy requirements this large force businesses to choose between the environment and maximizing capital, which rarely benefits the environment. 3

15 2.0 BACKGROUND Previous work in acid gas sweetening, the removal of carbon dioxide and/or sulfur dioxide from a gaseous process stream, is extensive and growing. There are many methods that have potential in carbon dioxide capture. These methods include chemical absorption, physical adsorption, O 2 /CO 2 recycle combustion, membranes, and cryogenic methods. Some methods use solid adsorbents, such as zeolites, and others use liquid absorbents. It is also possible to group methods together to increase performance. Typically, with the exception of the O 2 /CO 2 recycle system, the objective is to selectively separate CO 2 from a CO 2 -N 2 -O 2 -H 2 O mixture. 2.1 CRYOGENIC CAPTURE The cryogenic method can be used to directly liquefy a high purity carbon dioxide stream. This method provides a direct route for pipeline transportation to the sequestration point. 9 The cryogenic method requires the largest energy demands. This type of capture requires high carbon dioxide purity and has not been demonstrated on a technical scale. 10 The dryness and purity of the capture gas is important because large amounts of water may form ice or hydrates which could potentially cause pipe clogs. 11 This method also requires multiple compression and cooling stages to liquefy the carbon dioxide and remove other gases. 11 The energy demands are 4

16 between 0.6 and 1.0 kwh/kg CO 2 captured, which are at a minimum two times the energy demands of chemical absorption MEMBRANE CAPTURE Kaldis et al. 12 have shown that membranes offer possible carbon dioxide separation in flue gases. They studied a coal integrated gasification combined-cycle (IGCC) system that used a membrane to separate CO 2. The membranes that were studied were a polymeric membrane for a low temperature case and a porous ceramic membrane for high temperatures. The membranes were implemented in both single stage and multistage formations. Multistage formations increase final carbon dioxide purity but decrease final recovery substantially. Table 1 shows CO 2 recovery values for the two membranes. Table 1: CO 2 purity and recovery for membrane separation systems at 16 bar. 12 Polymer Membrane Ceramic Membrane One Stage Multi Stage One Stage Multi Stage CO 2 Purity (%) CO 2 Recovery (%) The ceramic membranes are far from being efficient enough for use, but the polymer membranes produce a relatively high carbon dioxide recovery. Membrane separation reduces efficiency of the plant by 8 14% depending on the pressure conditions and shift reactors. 12 The majority of the efficiency cost is due to compression of the gas streams. 11 This type of separation does reduce any hazardous chemicals or solvents that may be used in other types of capture systems. 5

17 2.3 O 2 /CO 2 RECYCLE COMBUSTION SYSTEM CAPTURE Carbon dioxide can also be captured by using an O 2 /CO 2 recycle combustion system. In this case, O 2 rather than air is used for combustion, and CO 2 is recycled to moderate reactor temperatures. This system has been done by Singh et al. 13 In this study an MEA capture system was tested against a O 2 /CO 2 recycle combustion system. Both systems were tested under similar conditions using Hysys and Aspen Plus software. The O 2 /CO 2 recycle combustion system was fit with a low temperature flash (LTF) unit to increase CO 2 purity above 98%. This additional system is not necessary because the O 2 /CO 2 recycle combustion system produces a maximum purity of 95% CO 2. This system is based on the combustion of pure oxygen in the furnace instead of air. The oxygen is purified by cryogenic air separation. This combined with the recycle stream should remove other gases leaving highly pure carbon dioxide which can be directly captured from this stream. The simulations infer that both systems are expensive retrofits of the original plant. The MEA system costs 3.3 cents/kwh, while the O 2 /CO 2 recycle combustion system costs 2.4 cents/kwh. The difference in cost occurs because of the cost of amine chemicals (per year). The O 2 /CO 2 recycle combustion system requires the additional LTF to purify the carbon dioxide close to the purity produced by MEA systems, which are greater than 99%. 14 Also the MEA systems typically capture 75-90% of the total carbon dioxide in the flue gas. The O 2 /CO 2 recycle combustion system only captured 74% of the carbon dioxide emissions. This system may be less expensive but it is not as proficient in reducing CO 2 emissions. Chemical looping combustion 15 (CLC) is another alternative for power generation with carbon dioxide recycle and capture similar to O 2 /CO 2 recycle combustion systems. CLC utilizes oxidation and reduction reactions linked by a metallic oxygen carrier to exchange O 2 between the 6

18 oxidation and reduction reactions. A stream of carbon dioxide and water is produced as the output stream of the reduction reaction. This captured CO 2 -H 2 O stream can be used to preheat the fuel and generate some power using a CO 2 turbine. The water vapor can be separated from the carbon dioxide rich stream leaving a pure stream of captured CO 2 which can be liquefied, similarly to the O 2 /CO 2 recycle combustion systems, and sequestered. This type of power generation is more efficient than conventional combustion systems even when the CO 2 capture costs are included in the energy balance PHYSICAL ADSORPTION The physical adsorption of amines can be done using chemical solvents or solid support systems. Chemical solvents include sulfolane (TMS) 16,17 and n-methyl-2-pyrrolidone (PZ). 17 Solid porous materials include activated carbon 18,19,20, zeolites (13X) 18, and carbon molecular sieves (CMS). 19,20 Physical adsorbents require the lowest energy demands of all types of capture systems. Physical adsorbents only require 0.09 kwh/kg CO 2 when equal pressures in the product and feed are used. 10 This is much smaller than that of chemical absorbents (0.34 kwh/kg CO 2 ). 10 Liquid physical absorbing systems are not efficient compared to the solid adsorbers. These systems are much more effective when combined with amounts of chemical absorbent materials. Sulfolane and PZ systems are more effective when combined with small amounts of MEA (15%) not only in binding, but also in regeneration costs. 17 Also at partial pressures greater than 50 kpa of carbon dioxide, mixtures of PZ and MEA are more adsorbent than MEA and water but not at lower partial CO 2 pressurees. 17 High carbon dioxide concentrations in flue 7

19 gases are necessary for these processes to work as efficiently as some of the amine based systems. Yang et al. have done work with solid physical adsorbents. 18,19 They have done pressure swing adsorption methods with activated carbon, CMS, and zeolite 13X. Also Siriwardane et al. 20 have done studies on molecular sieves and activated carbon similar to those done by the Yang group. Yang proposed that zeolite 13X has more capacity for binding carbon dioxide than the activated carbon, 18 but binds carbon dioxide too tightly making adsorbent regeneration too energy expensive. 19 The same group also proposed that activated carbon was a better adsorbent than carbon molecular sieves. 19 Activated carbon has less diffusional resistances than CMS which allows it to bind more carbon dioxide and desorb it faster when the pressure is reduced. This work was supported by the Siriwardane group. 20 This group showed that molecular sieve 13X was more selective than activated carbon at low pressures towards CO 2 over N 2 and binds more CO 2, but activated carbon binds more carbon dioxide at high pressures. At pressures of 45 psia, both adsorbents bind to approximately 4.1 mol CO 2 /kg adsorbent. At pressures below this the molecular sieve binds more carbon dioxide. At high pressures carbon dioxide binds to activated carbon up to 8.8 mol CO 2 /kg adsorbent (275 psia), while sieve 13X only binds 5.2 mol CO 2 /kg adsorbent. 20 This is due to the higher surface area of the activated carbon, 897 m 2 /g, than the 13X, 506 m 2 /g. 20 Physical adsorbents are possibilities for carbon dioxide capture. These systems show high binding to carbon dioxide but the purity of the gas decreases significantly with recovery. Activated carbon is the best type of physical adsorbent demonstrated in the literature. This adsorbent only produces a 75-80% pure carbon dioxide stream at a recovery at only 90%. 19 MEA systems, as was stated earlier, have recoveries greater than 99% with much greater purity. 8

20 These systems do not require as much energy, but the recovery is far less than chemical absorption methods. 2.5 CHEMICAL ABSORPTION Chemical absorption of carbon dioxide is the most studied method of carbon dioxide capture. 10 The most common chemical systems are liquid systems, but some work on solid adsorbents has been done. Solid adsorbents use amine groups bound or placed on support systems. Work has been done showing that high molecular weight amines can be loaded onto carbon nanotubes. 21 Carbon dioxide can also bind to primary and secondary amines tethered to silica 22,23 These systems have only been shown to adsorb carbon dioxide but binding is not efficient for use in industrial applications. Diaf and Beckman used polymers to bind carbon dioxide Diaf, Beckman, and Enick 27 also showed that amine functionalized polystyrene can be foamed by carbon dioxide. Seckin et al. 28 created polymers containing amidine compounds that bound carbon dioxide. R 1 NH 2 + CO 2 R 1 H 2 N CO 2 R 1 H 2 N CO 2 + R 2 NH 2 R 1 H N CO2 H 3 N R 2 Figure 1: Primary and secondary amine systems bind CO 2 at a 2:1 ratio. 9

21 In 1968, Caplow s 29 proposal that a carbamate was formed when carbon dioxide reacted with amines was broadened by Danckwerts 30 who stated that the mechanism was base catalyzed, not just by more amine, but by any base in solution with the amine. In this case the base removes the proton from the zwitterion. Other authors, such as Versteeg 31, have done extensive studies on reaction orders and kinetics of different primary and secondary amines. In this case carbon dioxide reacts with amine, then a second amine deprotonates the first complex. The second amine is required for the carbon dioxide to bind. This limits the capture to a 2:1 amine to carbon dioxide binding ratio. The mechanism is shown in Figure 1. This reaction mechanism is the basis for carbon dioxide capture systems being used today. There are many amine compounds being studied that exhibit this binding. Monoethanolamine (MEA), diethanolamine (DEA), diisopropanolamine (DIPA), triethanolamine (TEA), and n-methyldiethanolamine (MDEA) are common carbon dioxide adsorbing materials that have been extensively studied in previous work. MDEA and TEA are actually tertiary amines, but they are not used as commonly as the other compounds. Usually they are used as the base in mixed amine systems. 32,33 Mixed amine systems have been used in liquid amine systems. These mixed amine systems typically consist of a primary or secondary amine with a tertiary amine. 31 Mixed amine systems are used to take advantage of properties unique to each type of amine used. Tertiary amines react more slowly with carbon dioxide than the primary or secondary amine system. These mixtures are useful because the tertiary amines assist in carbon dioxide binding while regenerating faster during the desorption process. Primary and secondary amines in the mixture help increase the reaction rate. Glasscock and Critchfield 32 have shown that systems consisting of MEA/MDEA and DEA/MDEA absorb and desorb carbon dioxide. The MEA system is the 10

22 better system. MEA systems are used quite frequently in industry. 7,13 This allows for tailoring of gas streams to the specifications of a facility based on the amount of carbon dioxide. Table 2: Structures and pka values of common amines Chemical Name MEA HO Structure NH 2 Molecular Weight (g/mol) pka Reference Laddha and Danckwerts 8 DEA HO H N OH Laddha and Danckwerts 8 OH DIPA H N OH Blauwhoff, Versteeg, and Van Swaaij 9 MDEA HO N OH Blauwhoff, Versteeg, and Van Swaaij 9 TEA HO HO N OH Blauwhoff, Versteeg, and Van Swaaij 9 Sartori 33 has been a part of developing the Exxon Mobil based FLEXSORB absorbents. Sartori has used hindered amine systems, which include amines such as 2-amino-2-methyl-1-propanol, 1,8-p-menthanediamine, and 2-isopropylaminoethanol. Figure 2 is an example of the carbonate reaction with 2-amino-2-methyl-1-propanol and carbon dioxide in the presence of water. These hindered amine systems are unable to form carbamates which increase their binding capacity to carbon dioxide. Binding of hindered amines occurs at a 1:1 ratio of amine to gas. Hindered 11

23 amines benefit in binding as well as release of carbon dioxide. The carbamates are very stable which prevents all the carbon dioxide from being released while the hindered method of carbonate formation is fully reversible. The FLEXSORB PS system was developed by Sartori 33 and is used as a non-selective adsorbent that removes both H 2 S and CO 2. This uses a moderately hindered amine in an organic solution. The solution also consists of water. 33 This amine system, when compared to a conventional amine system, produced 125% more carbon dioxide removal at half the flow rate. 33 The system also only used 50% of the steam that the conventional primary or secondary amine liquid system uses. As mentioned before, 80% of the energy used in a capture process is due to the steam requirements. Cutting this value in half greatly improves efficiency. HO HO C NH 2 + CO 2 + H 2 O C NH 3 HCO 3 Figure 2: Hindered amines, such as 2-amino-2-methyl-1-propanol, bind at a 1:1 ratio. 33 Polymer systems containing amines have been shown to bind carbon dioxide. Diaf, Beckman and Enick 27 combined a styrene vinylbenzylchloride (VBC) copolymer with ethylenediamine (EDA). The polymer does reversibly bind carbon dioxide and can be regenerated above 60 ºC. 24 This discovery led to further studies on carbon dioxide adsorption in solid polymer systems. Diaf and Beckman 25,26 used styrene-vbc copolymers to bind carbon dioxide and other acid gases. They used primary, secondary, and tertiary amine based polymer systems with the most basic being primary amines and the least basic tertiary amines. 25 Carbon dioxide binds more strongly to the more basic amines which made EDA the most efficient amine compound for the polymer systems. 26 A copolymer containing the maximum amount of EDA 12

24 was shown to bind 2.5 mol CO 2 /kg polymer. 26 This number is smaller than typical values shown by others mentioned above. Jessop et al. 34,35 proved that 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) bound to carbon dioxide following the mechanism introduced by Sartori, that a hindered amine forms a carbonate anion system. Some confusion existed under which mechanism amidine compounds bound carbon dioxide. Jessop et al. 34 showed that DBU does not bind carbon dioxide when water is not present, and proved by conductivity experimentation, that DBU reacts with the carbonate anion in the presence of water. If zwitterions were present, there would be no additional conductive ions in the solution, but there was an increase in conductivity of the solution. 34 There is no evidence that the zwitterions forms between CO 2 and DBU when no water is present. This verifies that amidine compounds behave like hindered amine systems. Amidine compounds, like DBU, are highly basic and have high pka values that enable them to bind carbon dioxide tightly. Figure 3 depicts the carbonate reaction of DBU with carbon dioxide at a 1:1 ratio. N N H 2 O CO 2 N NH + - HCO 3 Figure 3: DBU binds CO 2 at a 1:1 ratio. No carbamate is formed by this mechanism. 34 Carbon dioxide capture in polymer systems that are more comparable to other forms of capture was achieved by Seckin et al. 28 that used polymers containing pendant 1,4,5,6- tetrahydropyrimidine polymer systems. These systems bound 3.4 mol CO 2 /kg polymer. 28 The increase in carbon dioxide binding is due to the 1:1 nature of amidine groups in the 1,4,5,6-13

25 tetrahydropyrimidine. 34 Due to the promise of this amidine polymer system, more research is necessary to explore carbon dioxide binding with other amidine systems. 14

26 3.0 HYPOTHESIS This research has been conducted based on the hypothesis that amidine groups are ideal tertiary amine systems for carbon dioxide capture in a solid adsorbent system. Amidine functional solid polymers and adsorbents are the focus of this research, because they allow for high capture capacity compared to conventional liquid systems while reducing the energy requirements needed for regeneration and handling. Amidine compounds exhibit a 1:1 binding ratio as discussed earlier when carbon dioxide is present with water as shown by Jessop et al. 34 Compounds containing these groups are more basic than typical amine groups. Carbon dioxide forms carbonic acid in water while the basic amidine groups protonate. The protonated amidine groups, such as DBU, can bind tightly to carbonate anions in aqueous solutions, as shown in Figure 3. This reaction can theoretically take place with one water molecule and one carbon dioxide present per each amidine group. 34 This one-to-one binding provides for a twice-asefficient adsorbent compared to primary and secondary amine systems. Imine groups may seem like the logical choice as a tertiary amine group, but imines are very unstable in water. 36,37 The differences between these tertiary imine systems are shown in Figure 4. 15

27 N R 1 R 6 Imine R 5 H N N R 1 N R 2 N R 1 N R 2 R 3 Amidine Guanidine R 3 Figure 4: Imines are unstable while amidine and guanidine groups are stable. 3.1 BINDING EFFICIENCY Although it is important increase the binding efficiency of carbon dioxide adsorbents, there are other factors that need to be taken into account in order to improve the overall binding efficiency. Systems today have high heat capacities because they are typically liquid systems that contain water, so the reduction of heat capacity will relieve the energy intensive regeneration process. If the adsorbents were solid systems with adsorbing groups bound to the surface then there would be a lower heat capacity to overcome allowing for rapid regeneration with minimal drain on the overall efficiency of the plant. For example an MEA/MDEA based system containing 80% MEA has a heat capacity of 182 J/mol*K (30ºC) 38 compared to 75 J/mol*K of water at 25ºC. A solid porous silica has a heat capacity of 42.2 J/mol*K, 39 less than 25% of the MEA/MDEA liquid system. In order to be industrially useful, a minimum binding of 3 mol CO 2 /kg adsorbent is preferred based on the following information. The maximum amount an aqueous 30 wt% MEA solution can bind is approximately 2.5 mol CO 2 /kg absorbent, and a 50 wt% MDEA solution can bind just over 4 mol CO 2 /kg absorbent. 7 Also MDEA is not used as 16

28 the sole absorbent, but in mixtures with MEA or DEA. These mixtures have been cited to have a total amine content of 25 wt% which at best could bind half of the 50 wt% MDEA STABILITY Stability of the adsorbent is also another important factor for a complete adsorbent. These adsorbents will release carbon dioxide when the water begins to evaporate out from the adsorbent. Adsorption of carbon dioxide should occur between 25 ºC and 65 ºC, because feed streams into typical amine systems are around 40 ºC. 13 Carbon dioxide desorption occurs near 100 ºC. 23 These systems must also be structurally stable in the hot environment of regeneration. Another important factor in determining the potential of a carbon dioxide adsorbent is the volatility of the compound. Amine adsorbents must have a low or negligible volatility so they do not evaporate. This will allow any liquid adsorbents that may be coated on a surface to be reusable without loss in the regeneration process. By minimizing volatility, amidine systems will be more robust in various temperatures. 3.3 SPECIFIC PROJECT AIMS This project focuses on carbon dioxide capture using solid amidine functional adsorbent systems. Solid CO 2 adsorbent systems allow for efficient carbon dioxide capture and release. A major focus of the project is to create efficient capture systems that can compete with current liquid carbon dioxide capture systems. Reducing superfluous weight and minimizing the equivalent 17

29 weight of the adsorbent allows for competitive binding efficiency. The following list the specific aims of this research: Synthesis and characterization of supported amidine and guanidine compounds, including amidine and guanidine solid polymer systems, which allows for efficient binding with reduced degradation. Reaching a capacity of 3 mol CO 2 /kg polymer or more in order to be competitive with liquid amine systems. Examine carbon dioxide binding of potential adsorbents, including regeneration of the carbon dioxide adsorbents. These adsorbents must be regenerable in order to be economical. Also binding should occur at low temperatures, ºC, with release approaching 100 ºC. Examine structural effects of the polymer and amine component on carbon dioxide capture in amidine and guanidine compounds. 18

30 4.0 EXPERIMENTAL The triethylene glycol diamine (XTJ-504) was purchased from Huntsman and used as received. All other reagents purchased in this research were purchased from Aldrich. They were used as received, without further purification. Butyllithium and DBU were kept refrigerated. Infra-red spectra were obtained on a Matson FT-IR on NaCl disks. A Brücker 300MSL was used to gather 1 H-NMR spectra. The solvent used for the NMR was CdCl 3 except in the case of the polyamidine when DMSO-d 6 was the solvent. The carbon dioxide and nitrogen gases used in the batch reactor were purchased from Penn Oxygen & Supply Co. 4.1 PREPARATION OF 1,8-DIAZABICYCLO[5.4.0]UNDEC-7-ENE (DBU) Due to the strong basicity of DBU, a good starting point is to functionalize DBU and use it as a first candidate for binding in this research. DBU is easily obtained and easy to work with as a nonviscous liquid, and DBU has been shown to be an efficient CO 2 binder by binding at a oneto-one ratio due to the inherent amidine group in the bicyclic structure. 34 DBU must have a proton removed in order to create an attachment point onto a solid system. DBU was deprotonated by the following method of Tomoi et al. 41 Add 125 ml THF to a round bottom flask and flush with argon. Add 5 mmol (760 mg) of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) to a round bottom flask. Cool in a dry ice/acetone bath while stirring. Add 4.75 mmol (5.0 ml) 19

31 1.0 M butyl lithium (BuLi) in hexane slowly over 30 minutes. Then while stirring for another 30 minutes slowly add the other desired halogenated reactant; 5 mmol 3-chloromethylphenylsilica, 5 mmol of 1-bromododecane, or 5 mmol of 1,4-dibromobutane. Remove from the dry ice/acetone bath and allow the solution to cool to room temperature while stirring for an additional 3 hours. Add 2 ml of methanol to the reaction. N BuLi -78 C N DBU N Li N Figure 5: Proton removal on DBU by BuLi 4.2 DBU SILICA DBU silica was prepared because it is a free flowing bead system that is easily heated to regenerate the original adsorbent. In order to create a DBU functionalized silica system, the DBU must undergo two steps. The first step is to create a link between the DBU and the silica by adding (p-chloromethyl)phenyltrimethoxysilane. Then the DBU is bound to the silica. This reduces any volatility of the DBU and allows for much simpler handling of the adsorbent p-chloromethylphenyl silica A procedure derived from Leal et al. 23 was used in this synthesis. Add 2.0 g of silica gel, purchased from Aldrich, to 100 ml of toluene in a round bottom flask. Then add 10 ml of (pchloromethyl)phenyltrimethoxysilane, purchased from Gelest, to the reaction. Reflux between 20

32 65-70 ºC for 18 hours. Filter product and wash with toluene and pentane. Allow the product to dry DBU-methylphenyl silica 5 mmol of DBU and 4.75 mmol of BuLi were combined in a round bottom flask under an inert atmosphere at -72 ºC as described above. Also added to the reaction mixture was the p- chloromethylphenyl silica mentioned above which had a weight of g. The reaction was stirred over night at room temperature. Add 7 ml of methanol to quench reaction. The powder has a slight yellow color and can be washed with THF to remove excess DBU and BuLi. The only characterization performed on the product was by TGA under CO 2 atmosphere. The product yield can be estimated by total amount of carbon dioxide captured vs. the total capture capacity. In this case a 43.4% yield was estimated (Section 5.1). Cl DBU O Si O O O Si O O Figure 6: Schematic of the DBU-methylphenyl silica synthesis. 21

33 4.3 1-DBUDODECANE Another method of reducing the volatility of DBU is to increase its molecular weight, as in the creation of 1-DBUdodecane. By adding a twelve carbon chain to DBU, the molecular weight of the absorbent is doubled. This greatly reduces the loss of absorbent during regeneration. Add 10.5 mmol (1.599 g) DBU and 25 ml THF to a round bottom flask in Ar atmosphere in a dry ice/acetone bath. Then, slowly, over a period of 30 minutes add 10 mmol 1.6 M BuLi to the reactor. Then 10 mmol (2.492 g) of 1-bromododecane was added to the reaction slowly over 30 minutes and stirred for an additional 30 minutes in the cold bath. The cold bath was removed and the reaction was allowed to continue overnight. This procedure was also adapted from Tomoi et al. 41 The reaction was washed with hexane and centrifuged in order to precipitate and remove LiBr. The solvent was evaporated leaving the product. The reaction produced a white liquid and had a yield of 66.6%. In order to reduce the volatility of DBU an alkyl chain of chain length of twelve was attached to the DBU. 1 H-NMR (CdCl 3 ), δ (ppm): 0.83 (-CH 3, 3H), 1.23 (- CH 2 -, 31H), 1.83 (-CH 2 CH 2 Br, 2H), 2.05 (-C=NCH 2 -, 2H), 3.23 (-NCH 2 -, 4H), 3.40 (-CH 2 Br, 2H). N BuLi -78 C N DBU N Li N DBU + Br DBU Figure 7: 1-DBUdodecane reaction scheme 22

34 4.4 1,4-DIDBUBUTANE In order to minimize the equivalent weight of the adsorbent and keep low volatilities, 1,4- didbubutane was synthesized. This compound doubles the capacity for CO 2 capture while maintaining a similar molecular weight to that of 1-DBUdodecane. DBU was deprotonated as described above and combined with 1,4-dibromobutane at a 2:1 ratio. A 10% excess of DBU was used in this reaction. The reaction was carried out in similar fashion to that of 1- DBUdodecane. 41 Add 25 ml THF and mol (3.141 ml) of DBU to a flask under Ar in a dry ice/acetone bath. Then slowly drip in 0.02 mol (12.5 ml) of BuLi over 30 minutes. Add 0.01 mol (2.16 g) of 1,4-dibromobutane to the reaction and stir of an additional 30 min. Remove from the cold bath and stir at room temperature for an additional 3 hr. The product in this case was a yellow gel-like solid. In order to purify, the product was dissolved in benzene and the LiBr that precipitated out was removed by centrifuging the reaction solution dissolved in benzene. Excess benzene was evaporated leaving the 1,4-diDBU butane. A yield of 93% was observed. 1 H-NMR (CdCl 3 ), δ (ppm): (-CH 2 -, 20H), (=N-CH 2 -, 4H), (- CH 2 CH 2 Br), (-N-CH 2 -, 8H), (-CH 2 Br), (H 2 O), (H 2 CO 3 ). 4.5 POLYETHER DBU Due to the difficulty in purifying the high molecular weight DBU analogues, polymer synthesis was the next logical step. Polymer chemistry allows for multiple amidine groups within a short period of one another while eliminating concerns of volatility. Polyether DBU was considered 23

35 as an elegant method of attaching DBU to a polymer structure that reduces the equivalent weight of the polymer. Polyepichlorohydrin was purchased from Aldrich and used without any further purification. A 5 % excess of DBU (1.697 ml) was deprotonated using the procedure outlined in the activation of DBU in a minimal amount of THF. After the 10.8 mmol (6.75 ml) of 1.6 M BuLi was added to the reaction, a solution of 1 g polyepichlorohydrin that was previously dissolved in 25 ml THF under Ar atmosphere, was added by syringe. The solution was stirred for an additional 30 minutes and removed from the cold bath. The solution was then stirred overnight at room temperature. After the completion of the reaction, excess solvent was evaporated and the polymer was allowed to air dry. Then slight heat was applied to remove any water and bound CO 2. The overall reaction yield was 25.2%. The product appeared as a dark yellow or brown solid and was easily elongated but not elastic, unlike the original polyepichlorohydrin. Due to the physical properties of the polyether, attempts to crosslink the polymer were made. Cross-linking the polymer would ideally allow for a more stiff structure, reducing the elongation that the current polyether displays. The polymer was cross-linked with 1,3- diaminopropane. This was done by adding 1 g of polyepichlorohydrin to a reaction mixture of 15% 1,3-diaminopropane and 70% DBU. This reaction resulted in a white amorphous solid that was less adhesive than the original product but exhibited similar elastic properties to the original polyether DBU. Further cross-linking will only further hamper the adsorbance of the polymer so no further syntheses were attempted. Carbon dioxide capture is reduced by increasing crosslinking sites. By reducing the binding by further than 70%, as seen above, removes any economic feasibility of the polyether DBU. 1 H-NMR (CdCl 3 ), δ (ppm): 1.41 (-CH 2 -, 8H),

36 (THF solvent) 2.25 (-C=NCH 2 -, 2H), 3.62 (-NCH 2 -, 4H), 3.69 (-CH 2 O-, 4H), 4.03 (-CH 2 Cl, 2H), 4.98 (H 2 O). * * O n + DBU * * O n Cl DBU Figure 8: Polyether DBU reaction scheme. DBU forms a bulky side chain in the polyether. 4.6 POLYGUANIDINE In order to test the effectiveness of binding between amidine and guanidine groups a polymer consisting of guanidine groups in the main chain was produced following the procedure outlined by Feiertag et al. 42 This polymer also minimizes the equivalent weight of the adsorbent. Combine 3.1 g ( mol) triethylene glycol diamine (XTJ-504) and 2 g ( mol) guanidine hydrochloride in a round bottom flask. The components were added at a 1:1 molar ratio of guanidine hydrochloride to triethylene glycol diamine and stirred while increasing the temperature to 160 ºC within 30 minutes. The reaction was refluxed overnight while evolving ammonia. The ammonia was neutralized by passing through a solution of HCl. After the reaction was completed, the polyguanidine was allowed to cool to room temperature. The reaction turned from a clear, nonviscous liquid to a brown, solid. The solid was stringy and amorphous. The final product weight was g producing a yield of 94%. 1 H-NMR (CdCl 3 ), δ (ppm): 1.12 (-CH 2 NH 2, 2H), 1.25 (-NH, 3H), 1.92 (-CH 2 NH-, 4H), 3.44 (-CH 2 O-, 8H). 25

37 NH H 2 N NH 2 HCl + H 2 N O O NH 2 NH R 1 N H N H O O n R 2 + HCl Figure 9: Polyguanidine reaction scheme. 4.7 POLYAMIDINE Similarly to the polyguanidine above, a straight chain polymer consisting of an amidine group in the backbone can be produced. This was produced using the same principles as the other synthesized polymers with regards to minimizing the equivalent weight of the adsorbents. The small repeat unit on the polyamidine creates an ideal adsorbent because many binding sites are located in close proximity to one another while minimizing extraneous polymer weight. The procedure by Böhme et al. 43 was followed. The diamine used was changed to a short alkyl chain in order to increase the basicity and maximize the reactive groups per total weight of the adsorbent. A distillation system was set up and flushed with nitrogen gas. Add 50 mmol (9.12 ml) of triethylorthoacetate to the three-neck round bottom and the temperature of was increased to 85 ºC. Then 47.5 mmol (3.97 ml) of 1,3-diaminopropane was added to the flask ml of hydrochloric acid (37%) was added slowly to catalyze the reaction. The reaction was allowed to proceed for an hour and then the temperature was increased to 120 ºC and another 0.5 ml of HCl was added. The reaction proceeded for an additional hour and the temperature was increased to ºC. A vacuum was slowly drawn on the reaction to induce the distillation 26

38 of ethanol. The reaction was carried out for another two hours under vacuum and when complete allowed to cool to room temperature and filled with nitrogen. No further purification was done to the polymer because all reactants are volatile and boil at temperatures under 160 ºC, so all impurities were distilled out of the polymer during the reaction. It is important to pull the vacuum slowly only to reduce pressure on the reaction before completely minimizing the pressure. If the vacuum is pulled to quickly the polyamidine consists of short chains and will form a pink clear liquid and a white solid. When pulled slowly the ethanol distills less violently reducing the amount of reactant that is lost in the distillation process. There should be no visible reactant at the bottom of the distillation flask after the reaction. If done correctly the polymer appears as a waxy, light brown solid. There is typically a portion of the total yield that has a short chain length and remains a liquid. The total product yield is 89.8%, with 64.8% of the product in the solid form. The long chain polymer is preferred to the shorter chain. 1 H-NMR (CdCl 3 ), δ (ppm): 1.44 (-OCH 2 CH 3, 3H), 1.82 (-CH 2 -, 2H), 2.10 (-CH 3, 3H), 2.60 (-NH 2, 2H), 3.00 (-OCH 2, 2H), 3.26 (-NCH 2 -, 4H). H 2 N NH 2 + O O O * N N H * + OH Figure 10: Polyamidine reaction scheme. 27

39 4.8 CO 2 ADSORPTION EXPERIMENTAL Capture of CO 2 has been studied by thermogravimetric analysis (TGA), a pressurized batch reactor, and a packed bed reactor attached to a mass spectrometer. The TGA was used as a screening tool to identify possible CO 2 adsorbent compounds, similar to DBU. 1,5,7- triazabicyclo[4.4.0]dec-5-ene (TBD), and n-methylimidazole were two compounds screened that showed some CO 2 affinity. The samples were allowed to sit at room temperature and then placed in the TGA. The temperature was ramped from 20 ºC to 100 ºC at 2 ºC/min. These samples were held at 100 ºC for 30 minutes and the total CO 2 weight loss was calculated from the change in the sample weight. Samples were also tested for CO 2 adsorbance in a batch reactor. The reactor was a 10 ml stainless steel reactor that was modified with a pressure gauge. The inlet was pure CO 2 or nitrogen gas. Samples were screened with nitrogen only as a control. The samples were added to a sample vial. Water, in liquid form, was also added to the reactor and the reactor was flushed three times with the gas and then pressurized to 45 psi for 18 hours. The sample was weighed before being placed in the reactor and then again after the reaction was completed. The sample was then heated under nitrogen in a boiling water bath to remove all CO 2 and water. Then the amount of CO 2 was calculated in moles of carbon dioxide per kilogram of adsorbent. 28

40 5.0 RESULTS Results are based on the project aims that were discussed in the hypothesis section. Carbon dioxide adsorbents were synthesized and characterized according to their amidine or guanidine functionality. Then the synthesized adsorbents were tested for carbon dioxide binding capacity. Binding and structure were compared to examine the affects of structure on carbon dioxide capture. Amidine containing compounds were screened by TGA for weight gain and reversibility of carbon dioxide. The primary starting point of this research was DBU. This compound is a bicyclic amidine that binds strongly to carbon dioxide. 34 Amidine functional compounds, similar to DBU, bind well to carbon dioxide because they have high pka values. Table 3 shows the pka of several amidine or guanidine functional compounds. These pka values promote strong binding to carbon dioxide. DBU is a good choice for carbon dioxide capture because it has a pka value of The pair of nitrogen atoms separated by a double bond help donate electron density to the amine increasing the basicity of DBU. The downside to this increased basicity is the size of the molecule. Larger molecules add to the overall adsorbent weight, reducing the maximum amount of carbon dioxide capture per weight. Therefore, if the same or a similar pka can be kept while reducing the ineffective weight of the compound, the amount of carbon dioxide captured will increase. This is why low repeat unit molecular weight polyamidines and polyguanidines may be the best choices for carbon dioxide capture. 29

41 Table 3: pka Values of Amidine and Guanidine Functional Compounds. *pka of guanidine Molecular pka Chemical Name Structure Weight (Reference) (g/mol) 1,8- diazabicyclo[5,4,0]undec- 7-ene (DBU) N N (44) Imidazole HN N (45) 1,5,7- triazabicyclo[4,4,0]dec-5- ene (TBD) N N N H (44) Polyguanidine * (45) Polyamidine Up to 11.5 (46) 5.1 DBU-METHYLPHENYL SILICA The DBU-methylphenyl silica was synthesized as a preliminary attempt to bind carbon dioxide. This method was done using Merck silica gel which had a surface area of 750 m 2 /g, provided by Aldrich. Preliminary tests were done by TGA. When considering porous solid substrate carbon dioxide capture systems, it is helpful to estimate the theoretical maximum for carbon dioxide capture. Using common conversions combined with the area, binding ratio, and the amine surface coverage, the maximum carbon dioxide capture can be estimated. Using these simple calculations, the feasibility of a solid system can be directly estimated. Maximum carbon 30

42 dioxide capture by solid supported amidine functional compounds can be calculated using the following relations: A m 2 /g * (10 10 ) 2 (Å) 2 /m 2 = A (Å) 2 /g, (10 20 A (Å) 2 /g) / Z (Å) 2 /amine = A/Z amine/g, A/Z amine/g * R CO 2 /amine = AR/Z CO 2 /g, AR/Z CO 2 /g * 10 3 g/kg *1 mol/6.02x10 23 CO 2 molecules = AR/(6.02)Z mol/kg where A = area (m 2 /g), Z = surface coverage of amine (Å 2 /amine) R= binding ratio (CO 2 /amine) The maximum possible carbon dioxide capture of this type of adsorbent is based on the surface area of the silica, the binding ratio of the amine, 1:1 in amidine and guanidine cases; 34 and the surface area taken up by each attached amine unit. A generous assumption for an attached surface area would be 25 Å 2 /amine. A typical carbon-carbon bond length is 1.54 Å, therefore, when thinking about the area taken up by an attached molecule on a surface the area cannot be much smaller than the general assumption above. 47 Using this area and the known surface area of the silica the maximum amount of carbon dioxide that can be bound is 4.98 mol CO 2 /kg adsorbent. This is not promising, but the TGA analysis does show that the DBU silica does bind carbon dioxide at 2.16 mol CO 2 /kg adsorbent. This can be seen in Figure 11. The total weight loss of the sample was 11.8%, while 8.3% of the weight lost was from carbon dioxide. (The initial sample weight was mg.) 31